Institutional Sponsor

  • Black Twitter Icon

© 2019 by SciPy.

Giordon Stark
Lukas Heinrich
Matthew Feickert

pyhf: A pure Python Statistical Fitting Library from the High Energy Physics Community with Tensors and Autograd

In experimental high energy physics statistical fitting tools have been implemented almost entirely in C++. pyhf is a pure-Python implementation of the "HistFactory" statistical model, used to discover the Higgs boson, for multi-bin histogram-based analysis with interval estimation based on "Asymptotic formulae for likelihood-based tests of new physics" [arxiv:1007.1727]. pyhf supports modern computational graph libraries such as TensorFlow and PyTorch as computational backends to make use of features such as auto-differentiation and GPU acceleration. pyhf demonstrates a flexible and robust model for large binned statistical fits to meet the statistical challenges of the high energy physics community.